In designing self-administered paper questionnaires to obtain information about persons, two designs are frequently used: 1) A grid design with the questions along one axis and the person names along the opposite axis, and 2) a design with questions grouped together for each person (grouped questions design). The Census Bureau recently had an opportunity to test these two design options as part of a cognitive interviewing project on the Census of Juvenile Detention, Correctional, and Shelter Facilities. This census is conducted biennially at facilities (establishments) that house youth who are placed, committed or detained through legal or social service jurisdiction. Through cognitive interviews with 18 establishments, the Census Bureau investigated the use of the grid and grouped questions designs for collection of individual data on the children housed at these facilities. This paper describes outcomes of these interviews. Covered is a discussion of the perceived advantages and disadvantages of each design, and under what conditions one design is preferred over the other, e.g., number of children to be reported, use of paper vs. computer records, and whether the reporter provided the information for one or several facilities. Implications of these findings for data collection of person data for establishments will also be discussed.